Clinical, cytogenetic, and molecular insights from 32 years of the Portuguese Fanconi anemia cohort

Abstract

Purpose

Fanconi anemia (FA) is a rare monogenic chromosome breakage syndrome that presents with variable morphologic abnormalities and progressive bone marrow failure. Based on over 30 years of diagnostic experience, our main goal was to describe the Portuguese FA population and assess potential clinical and analytical phenotypic correlations with both molecular variants and chromosome instability (CI).

Methods

We evaluated the in-house database from the Cytogenetics Laboratory of the School of Medicine and Biomedical Sciences, University of Porto, which has provided nationwide CI testing since 1993.

Results

Ninety-three FA cases were diagnosed between January 1993 and October 2025; 1 was diagnosed in the prenatal setting due to severe malformations. Molecular characterization was available for 50 of these cases. Of these, 45 carried FANCA variants in homozygosity or compound heterozygosity, with 38% (of Romani ethnicity) harboring the homozygous NM_000135.4:c.295C>T p.(Gln99Ter) founder variant. Five additional cases carried variants in FANCD2, FANCG, FANCJ, or PALB2. The most common clinical presentation comprised both hematologic and morphologic phenotypes, observed in 47% of cases. Isolated hematologic and morphologic presentations were observed in 27% and 13% of cases, respectively, while the remaining 13% of patients were asymptomatic.

Cytogenetic diagnosis using the standard 1,2,3,4-diepoxybutane test was extremely accurate. Follow-up cytogenetic assessments revealed likely somatic genetic reversals in 2 cases occurring years after the diagnosis, suggesting the importance of reevaluating the 1,2,3,4-diepoxybutane test. CI correlated significantly with the hematologic phenotype and red blood cell analytic parameters.

Conclusion

Our data allowed the characterization of the Portuguese FA cohort from clinical, cytogenetic, and molecular perspectives, including the identification of population-specific molecular findings and associations between CI and clinical features. Findings related to cytogenetic assessment proved useful not only for diagnosis but also for FA follow-up and prognostic purposes.

Introduction

Fanconi anemia (FA) is a monogenic chromosome breakage syndrome caused by pathogenic variants in genes involved in a common pathway responsible for repairing DNA interstrand crosslinks.1, 2, 3 FA is genetically heterogeneous, as 23 genes (FANC-A to -X) have been implicated as causal.4, 5, 6, 7 Most genes are associated with an autosomal recessive mode of inheritance, while FA related to the FANCR/RAD51 (HGNC:9817) or FANCB (HGNC:3583) genes is associated with autosomal dominant or X-linked inheritance, respectively.^8,9 The most prevalent cases are attributed to variants in FANCA (HGNC:3582), followed by FANCC (HGNC:3584) and FANCG (HGNC:3588); variants in the remaining genes are considered rare causes of FA.^6,7,10

The estimated worldwide incidence of FA ranges between 1 in 250,000 and 1 in 100,000 newborns,¹¹ making it an ultra-rare disorder. However, there are specific populations where the incidence is much higher, such as the Iberian Romani and the Ashkenazi Jewish populations, likely due to the presence of elevated rates of consanguinity and founder variants.^12,13

FA is phenotypically heterogeneous. Its clinical presentation has been characterized primarily by morphologic and hematologic phenotypes and an increased cancer predisposition. Morphologic phenotypes are present in approximately 75% of patients, and most commonly comprise prenatal and/or postnatal short stature; skeletal malformations of the upper limbs, typically the thumbs; pigmentation anomalies such as café-au-lait macules; microcephaly; genitourinary malformations; and ophthalmologic anomalies.^6,7 The hematologic phenotype is typically progressive bone marrow failure, which more commonly arises during childhood but, in rare cases, can be absent until late adulthood or may not develop at all.^6,7,14 Chromosome instability (CI) in proliferating tissues increases the predisposition to both hematologic malignancies, particularly acute myeloid leukemia, and solid tumors, with head and neck squamous cell carcinomas and gynecologic cancers being among the most common.^3,7 Accordingly, some individuals are diagnosed with FA only following the development of malignancy.¹⁵ Additional functional manifestations, including hearing loss, endocrine dysfunction, and infertility, further contribute to the disorder’s broad clinical spectrum.^6,7

The diagnosis of FA can be established through cytogenetic and/or molecular approaches, although a combination of both is preferable. The cytogenetic hallmark of FA is a significant increase in CI upon exposure of primary lymphocytes to bifunctional alkylating agents that induce DNA interstrand crosslinks.¹⁶ The introduction of the 1,2,3,4-diepoxybutane (DEB) test by Auerbach et al¹⁷ represented a landmark in FA diagnosis, as it provided a standardized and reproducible method to quantify CI in peripheral blood lymphocytes. This assay is highly specific and sensitive, showing no overlap in the mean number of chromosome breaks per cell–the main discriminative parameter–between patients with FA and control populations.¹⁷ Its diagnostic reliability, however, depends not only on the quantitative assessment of breaks per cell but also on the accurate recognition of chromosomal aberrations, particularly complex radial figures, which are a hallmark of FA. Cases that present such aberrations and a mean number of breaks per cell above the laboratory-specific threshold for chromosome breakage are thus classified as FA. The analytical experience and interpretive precision of the cytogeneticist are therefore critical, particularly for detecting mosaicism, which may lead to false-negative results if unrecognized. The introduction of the chromosomal fragility index (CFI) further refined diagnostic accuracy and revealed associations between fragility patterns and clinical manifestations.¹⁸ On the other hand, molecular confirmation can be obtained through sequencing of FA-associated genes; however, in a small proportion of cases (approximately 5%), no pathogenic variants are identified despite a positive cytogenetic diagnosis, likely reflecting limitations of current molecular approaches, such as deep intronic variants, regulatory variants, structural rearrangements, or low-level somatic mosaicism.¹⁹ Both approaches are complementary: cytogenetic diagnosis remains the gold standard for the diagnosis of FA and can overcome the small percentage of variants missed by conventional molecular testing; in addition, identification of the causal molecular alteration is essential for specific genetic counseling and offering reproductive options.

Despite the large genetic heterogeneity, there are few well-established genotype-phenotype correlations. This is likely due both to the rarity of the condition and to population-specific genetic and environmental modifying factors. FANCA-null variants have been associated with more severe or earlier hematologic phenotypes than hypomorphic ones, but this association has been further disputed.^20,21 Specific FANCB and FANCC variants have also been linked to different phenotypic severities and congenital malformation incidence, although population-specific effects were also observed for these associations,^22,23 which is a limitation of genotype-phenotype correlation assessment in ultra-rare disorders. This complexity is further illustrated by biallelic pathogenic variants in FANCM (HGNC:23168), which lead to FA pathway dysfunction with cancer predisposition, chemotherapy hypersensitivity, and CI, yet lack bone marrow failure. In line with this distinct clinical presentation, FANCM-associated disease is often excluded from the classical FA spectrum, and the recent ClinGen Childhood, Adolescent, and Young Adult Cancer Predisposition Gene Curation Expert Panel classification framework categorizes it as an “FA-like” disorder.24, 25, 26 Nevertheless, a recent cohort study continues to support the biological inclusion of FANCM-associated disease within the broader FA spectrum, reinforcing the clinical relevance of recognizing these patients within an FA-oriented diagnostic and management framework.²⁷

Similarly, individuals with FA caused by pathogenic variants in downstream genes of the FA/BRCA pathway, such as RAD51, BRCA1 (HGNC:1100), BRCA2 (HGNC:1101), and PALB2 (HGNC:26144), tend to present with an atypical FA phenotype, displaying a broader and more variable overall clinical spectrum, and are frequently characterized by an absence or marked attenuation of hematologic manifestations, particularly bone marrow failure.^28,29 Consistent with this, all previously reported cases of RAD51-related FA, all of which were caused by heterozygous missense variants, have not manifested bone marrow failure , even in adulthood.³⁰ Notably, BRCA1, BRCA2, and PALB2 are also well-established cancer predisposition genes in heterozygotes, with an increased risk, particularly for breast and ovarian cancer.²⁸

It is currently not established whether the degree of CI in peripheral blood cells correlates with the phenotypic presentation and its severity. While several somatic cytogenetic changes are hallmarks of myelodysplasia or acute leukemia,³¹ 2 complications of which patients with FA are at risk, the association between the degree of CI and the severity and variability of the hematologic and/or morphologic phenotypes is largely unaddressed. One of the reasons behind this is the variability of the analytical approach and established cutoffs used to classify CI. Nevertheless, a recent study in a large Spanish FA cohort showed that patients with malformations had increased CI upon DEB exposure (compared with those without malformations), although no difference was seen when comparing those with different numbers of malformations.³² In the same study, patients who were characterized as having high versus low CI had a more severe hematologic phenotype and prognosis.³² However, allocation of patients to high or low instability groups was made based on the median CI observed in that cohort, which is cohort-specific and based on a statistical, rather than biological, rationale.

Since 1993, the diagnosis of FA across Portuguese hospital centers has been ensured by the Cytogenetics Laboratory of the School of Medicine and Biomedical Sciences of the University of Porto (ICBAS). This resource has allowed for a detailed, longitudinal characterization of the Portuguese population with FA over 32 years, providing one of the largest national cohorts of patients with FA in Southern Europe and offering a rare opportunity to study this disease at a population level.³³ This provides a comprehensive characterization at the clinical, molecular, and cytogenetic levels, explores potential clinical/analytical phenotypic correlations, and provides data regarding the technical optimization in CI assessment that raises the standards for precise FA diagnosis. Moreover, this nationwide approach has enhanced patient care, supported early detection, and served as a benchmark for rare disease management and registry development in Portugal.

Materials and Methods

Patient data collection and inclusion criteria

All patient data were obtained from the Cytogenetics Laboratory of ICBAS. Since the establishment of the cytogenetic diagnosis in 1993, demographic, clinical, and laboratory data from all patients for whom an evaluation of CI was requested have been stored in an internal database. This includes patients from Portuguese hospitals nationwide who were referred due to suspicion of or exclusion of FA. We reviewed and screened the database for patients with a final diagnosis of FA. Additional clinical (history and physical examination findings) and laboratory data from positive cases were obtained through electronic health records of patients that were provided by the referring physician. All information regarding hematologic parameters was obtained and refers to the time of diagnosis, unless otherwise stated. In this study, we included patients who fulfilled either cytogenetic diagnostic criteria or molecular diagnostic criteria for FA. Whenever possible, cases with a cytogenetic-only diagnosis were molecularly tested, and cases with a molecular-only diagnosis were cytogenetically tested.

Control samples were recruited from among volunteer blood donors at the Immunohematology and Transfusional Medicine Unit, Imunohemotherapy Service, Unidade Local de Saúde Santo António. Informed consent was obtained from all donors, and the samples were anonymized. CI data collected from these individuals were used exclusively as negative controls for the DEB test.

Establishment of cell cultures

From each individual, 2 to 3 mL of heparinized peripheral blood collected by venipuncture into heparinized tubes were sent to the laboratory. Whole blood cultures were established by adding 0.5 mL of whole blood to a final volume of 7 mL of Roswell Park Memorial Institute-1640 complete medium supplemented with 15% heat-inactivated fetal bovine serum, L-glutamine, and penicillin–streptomycin antibiotic solution (Gibco, Thermo Fisher Scientific). Cultures were stimulated with 5 μg/mL phytohemagglutinin (Gibco, Thermo Fisher Scientific) and incubated at 37 °C in a humidified atmosphere containing 5% CO₂. In parallel, cultures from healthy blood donors were processed under the same conditions and used as negative controls. In the appropriate cultures, DEB (Sigma-Aldrich) was added at a final concentration of 0.5 μg/mL (24 hours after culture initiation). In parallel, an extra culture exposed to 1.0 μg/mL DEB (twice the diagnostic concentration) was prepared as a technical quality control to confirm DEB activity and effective exposure. Given that DEB is a suspected carcinogen, all procedures were performed under appropriate safety precautions.

After 72 hours of culture, cells were incubated with 0.13 μg/mL Colcemid (Gibco, Thermo Fisher Scientific) for 1 hour, followed by hypotonic treatment with 0.075 M KCl (Sigma-Aldrich) and fixation in a 1:3 acetic acid:methanol solution (Fisher Chemical). Chromosome preparations were obtained by air-drying and were stained with 4% Giemsa (Sigma-Aldrich).

Cytogenetic evaluation of chromosome instability

CI was evaluated according to well-established laboratory protocols. Analysis was performed on 25 to 100 metaphases from each culture condition by 2 independent scorers and in a blinded fashion. To avoid bias in cell selection, consecutive metaphases that appeared intact with sufficiently well-defined chromosome morphology were selected. Each cell was evaluated for chromosome number as well as for the presence and type of aberrations. Achromatic lesions narrower than 1 chromatid were scored as gaps and were excluded from the calculation of CI frequencies. Achromatic areas wider than a chromatid were scored as breaks. Dicentric and ring chromosomes were scored as rearrangements with 2 breaks. Radial figures were scored as rearrangements with 2 or more breaks, depending on the number of chromosomes involved. Radial figures involving 3 or more chromosomes were scored as complex figures. Cells presenting extremely high levels of CI, with more than 20 breaks, and therefore not allowing accurate quantification of damage, were classified as pulverized. Although these cells were included in the percentage of aberrant cells, they were excluded from the calculation of the mean number of breaks per cell.

CI results were thus expressed as the percentage of aberrant cells and the mean number of breaks per cell. A case was considered DEB-positive when the mean number of breaks per cell was above the in-house diagnostic cutoff established for patients with FA and complex radial figures were present. The specific in-house discriminative CI thresholds used for diagnosis are described in detail in the Results section. In parallel, the CFI was calculated to identify possible mosaic FA cases. CFI was obtained by multiplying the percentage of aberrant cells by the mean number of breaks observed in cells with 2 or more breaks (CFI = [% aberrant cells] × [breaks / multi-aberrant cells]).

Molecular diagnosis of FA

Molecular testing was requested by referring clinicians and was performed in different external accredited commercial labs. All tests for probands were based on a next-generation sequencing panel that included FA-related genes. In cases with a positive family history (ie, not the proband of a family), targeted sequencing for the familial variant was requested instead. We reviewed all relevant single nucleotide variants and performed an independent classification using the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) version 3.0 classification criteria,³⁴ including subsequent ClinGen-recommended modifications. For copy number variants (CNVs) detected by next-generation sequencing, a subsequent targeted comparative genomic hybridization array was requested from an external commercial lab to confirm true positive variants. CNV classification was then performed using the ACMG/ClinGen technical recommendations.³⁵ Cases with 1 homozygous variant or 2 compound heterozygous variants that were classified as likely pathogenic or pathogenic were considered molecularly confirmed positive cases. VariantValidator (University of Manchester)^36,37 was used for validating sequence descriptors and for mapping sequences between transcript and genomic variants.³⁸

Statistical analysis

Continuous variables are reported by their mean and standard deviation, by their median and range, or, for CI parameters, by the interval between the 2.5th and 97.5th percentiles. Categorical variables are reported by their frequency and proportion. For inferential statistical testing, a 95% confidence interval was assumed, and all tests were two-tailed. The assumption of normality was made a priori, per the central limit theorem, for all tested continuous variables with a sample size over 30. Otherwise, continuous variables were tested for normality using the Shapiro-Wilk test, inspection of histograms and Q-Q plots, and evaluation of the values of skewness and kurtosis. An independent samples t test was used to compare differences between 2 means. Correlations between 2 continuous variables were obtained using Pearson’s correlation coefficient (r) or, if at least one of the variables was discrete, using Spearman’s correlation coefficient (ρ). Comparison of proportions between groups was performed using Fisher exact test. Appropriate effect size measures are reported for all tests (Hedge’s g for t test, with cutoffs of 0.2, 0.5, and 0.8 for small, medium, and large effects; Pearson’s r for correlation, with cutoffs of 0.1, 0.3, and 0.5 for small, medium, and large effects; Cramer V for Fisher exact test, with cutoffs of 0.1, 0.3, and 0.5 for small, medium, and large effects, respectively). Data analysis was performed using GraphPad Prism version 10.6.0 (GraphPad Software) and SPSS 27.0 (IBM Corp.).

Results

Sample demographics and incidence estimation

Throughout a period of 32 years (January 1993-October 2025), a total of 93 cases of FA were diagnosed in the Cytogenetics Laboratory of ICBAS, comprising 92 postnatal cases and 1 prenatal case. This represents an accurate approximation of all diagnosed FA cases with clinical follow-up in Portugal, as the laboratory provides the national reference service for FA diagnosis and receives samples from the whole country. Sample demographics are summarized in Table 1. The distribution between male and female sex is similar, with a small preponderance of females (0.7:1 male:female ratio), which differs from the previously reported small preponderance of males (1.2:1).¹⁰ The median age at diagnosis is 6 years, ranging from 0 to 35 years. More than half of the cases are of White ethnicity, followed by cases from the Romani population, which represent more than a third of the sample. This is in line with the well-described Iberian Romani founder effect for FA.¹² More than half of the cases were referred by hospitals in the northern area of the country. Consanguinity was present in 38 (41%) cases, of which 35 (92%) were from the Romani community.

Table 1.

Characterization of sample demographics and genetics

Variable	Descriptives (N = 93)
Sex
Female	54 (58%)
Male	39 (42%)
Age at diagnosisa	6 (0-35)
Ethnicity
White	52 (56%)
Romani	36 (39%)
African	4 (4%)
Asian	1 (1%)
Geographical region of referring hospital
North	53 (57%)
Center	27 (29%)
South	12 (13%)
Islands	1 (1%)
Consanguinity	38 (41%)
Positive family history	17 (18%)
Molecular diagnosis	50 (54%)
FANCA	45 (90%)
HGNC:3582
FANCD2	2 (4%)
HGNC:3585
FANCG	1 (2%)
HGNC:3588
FANCJ/BRIP1	1 (2%)
HGNC:20473
PALB2	1 (2%)
HGNC:26144
Allelism
Homozygous	41 (82%)
Compound Heterozygous	9 (18%)

Descriptive statistics are presented as frequency (proportion) or as median (minimum–maximum).

HGNC, HUGO Gene Nomenclature Committee.

^aSingle case was diagnosed in the prenatal setting.

Out of the 93 diagnosed cases, 85 (91%) were born in Portugal. Among these, 61 were born in the 1993-2024 period. The national birth registry from the Portuguese National Statistics Institute indicates that there were 3,177,002 live births in Portugal during that period.³⁹ Based on this data, we can estimate the incidence of FA to be approximately 1 in 50,000 newborns for that period. In comparison, previously reported estimates for FA incidence are between 1 in 250,000 and 1 in 100,000,¹¹ which corresponds to a predicted heterozygote frequency between 1 in 250 and 1 in 158 under Hardy-Weinberg equilibrium. Higher frequencies (below 1 in 100) have been reported in populations with a founder effect, such as the Iberian Romani and Ashkenazi Jewish groups. Out of the 61 aforementioned cases, 36 (59%) are of Romani ethnicity (Table 1), suggesting a substantial contribution of the Romani founder effect to the higher incidence and heterozygote frequency estimated in our study. No data regarding the number of annual births in the Portuguese Romani community are available, and therefore a specific incidence/heterozygote frequency cannot be calculated. Nevertheless, our calculated values may underestimate the true burden of FA in Portugal due to incomplete case ascertainment, as some patients may remain undiagnosed; although we receive samples from across the country, we cannot ensure that all hospitals refer suspected cases to our center.

Internal establishment of the discriminative CI parameters for the diagnosis of FA

To establish the cutoff values of the CI parameters used for the cytogenetic diagnosis of FA, we analyzed data from 88 of the 93 cases included in this study. The remaining postnatal cases were excluded because cytogenetic testing was not performed at the Cytogenetics Laboratory of ICBAS. The prenatal case was also excluded because only molecular testing was performed.

The diagnostic assessment was based on the evaluation of 4 CI parameters: the percentage of aberrant cells, the number of breaks per cell, the percentage of multi-aberrant cells, and CFI. For each parameter, values were calculated for both untreated and DEB-stimulated cultures (0.5 μg/mL) (Table 2).

Table 2.

Distribution of chromosome instability (CI) parameters in control and Fanconi Anemia (FA) blood samples after untreated and DEB-stimulated cultures

Parametera	Untreated		DEB 0.5 μg/mL
	Controls (N = 250)	FA (N = 88)	Controls (N = 250)	FA (N = 88)
% Aberrant cells	0-2	0-50	0-19	22-100
No breaks/cell	0.00-0.02	0.00-0.93	0.00-0.28	0.96-15.37
% Multi-aberrant cells	0	0-29	0-7	15-100
Chromosomal fragility index	NA	NA	0.00-41.40	151.70-1670.80

DEB, diepoxybutane; NA, not applicable.

^aDescriptive statistics are presented as 2.5^th to 97.5^th percentile ranges per group.

In untreated cultures, representing spontaneous CI, values were globally low and overlapped between groups. Among controls, the percentage of aberrant cells ranged from 0% to 2%, with no multi-aberrant cells observed, and the maximum number of breaks per cell was 0.02. FA cases showed a wider range of spontaneous instability, reaching up to 50% aberrant cells and 0.93 breaks per cell. As there were several overlapping values, this confirmed that CI values in untreated cultures cannot be used for diagnostic purposes.

After DEB exposure (0.5 μg/mL), a clear separation between groups was observed. In controls, the percentage of aberrant cells ranged from 0% to 19%, with up to 7% multi-aberrant cells and a maximum of 0.28 breaks per cell. In contrast, FA cases displayed a markedly higher response to DEB stimulation, with 22% to 100% aberrant cells, 15% to 100% multi-aberrant cells, and 0.96 to 15.37 breaks per cell. These nonoverlapping ranges clearly distinguished the control and populations with FA (Figure 1A and Table 2).

Figure 1.

Assessment of chromosome instability in FA cases. (A) Distribution of the breaks per cell (in logarithmic scale) in controls and FA cases. Cases in blue represent likely FA mosaics. The positivity cutoff is at 0.96 breaks per cell. The dashed line represents the median. (B) Absolute mean breaks per cell in the 15 cases with follow-up cytogenetic data available. Each dot represents an individual assessment at a specific age. Each colored dot represents an individual patient, with the exception of the cases with orange dots (2 siblings). Bold colored lines represent patients with likely somatic genetic reversal, while the gray line with orange dots represents the sibling of one of the patients. FA, Fanconi anemia.

CFI values in our cohort ranged from 0.00 to 41.40 in controls and from 151.70 to 1670.80 in FA cases, with no overlap between the 2 groups. The CFI values in the control population are in accordance with established reference ranges,¹⁸ while the lowest CFI observed among FA cases fell within the range defined for mosaic FA, consistent with the cases that were cytogenetically diagnosed as mosaic in this cohort.

In summary, based on the overall analysis, a diagnostic cutoff of ≥0.96 breaks per cell at a DEB concentration of 0.5 μg/mL was confirmed as the optimal maximum threshold to consider a DEB result positive, classifying individuals as having FA. This value aligns with the thresholds reported in previous studies¹⁷ and supports the robustness of the DEB test.

In our cohort of 88 individuals, 2 cases clinically or molecularly identified as FA presented CI values below the predetermined cutoff for diagnosis (Figure 1A).³⁵ These cases presented 0.68 and 0.86 breaks per cell, that is, intermediate values between FA and control values, and the CFI values were 151.70 and 178.56, respectively, which were classified as mosaic FA. Molecular confirmation at the time of diagnosis was not possible because genetic testing was not requested in 1 case, while in the other, exome sequencing analysis did not identify any pathogenic variants in FA genes. Nevertheless, the clinical phenotype and intermediate CI values are strongly indicative of a mosaic FA diagnosis.

Reevaluation of chromosome instability in clinical follow-up

For 15 of the cases, additional CI evaluations were performed during clinical follow-up. Reevaluations were conducted in the context of individual clinical evolution, including changes in the hematologic phenotype, suspicion of somatic mosaicism, or pre-transplant assessment. Although the timing of CI assessments varied considerably across cases, we observed that the number of breaks per cell was variable, without a consistent pattern over the years (Figure 1B). Despite this variability, in almost all cases, CI parameters remained above the diagnostic cutoff for FA, including those with an improvement in the hematologic phenotype.

Nevertheless, we observed 2 cases in which CI follow-up values decreased below the cutoff for FA diagnosis. Importantly, both cases had been diagnosed in childhood in a familial FA context, following the diagnosis of an affected sibling, and showed clearly positive CI results at diagnosis. The first case (full red line) had no congenital malformations or peripheral cytopenias at diagnosis, although mild macrocytosis was present. In adulthood, she developed manifestations within the broader FA clinical spectrum, including vulvar squamous cell carcinoma and premature ovarian insufficiency. Two reevaluations in adulthood revealed an almost complete absence of chromosome breakage in peripheral blood. The second case (full orange line) presented with several clinical features compatible with FA, including poor growth/failure to thrive, café-au-lait macules, mild thumb dysmorphism, congenital heart defects, and renal asymmetry with hypoplasia of the right kidney. Reevaluation in adolescence showed a complete absence of chromosome breakage in peripheral blood. Interestingly, her sibling (gray line with orange dots) was also reevaluated and, to date, has not shown a negative CI test result.

These 2 cases likely reflect the phenomenon of somatic genetic rescue, resulting from complete reversion of the pathogenic allele in hematopoietic cells, as previously reported in patients with FA.⁴⁰ Notably, both cases were asymptomatic from a hematologic point of view both at diagnosis and remained so at the time of follow-up cytogenetic testing, raising the possibility that hematopoietic somatic reversion may have contributed to their preserved hematologic phenotype.

Cytogenetic quantification and characterization of chromosome breakage in FA

In addition to the main CI parameters, we also analyzed the distribution and complexity of chromosome breakage observed in FA and control samples (Table 3). CI parameters were assessed in spontaneous and DEB-treated cultures and included the number of breaks and fragments (Figure 2G-H), the number of radial figures (Figure 2I-J), the number of complex radial figures, the number of ring chromosomes (Figure 2K-L), the number of dicentric chromosomes (Figure 2M-N), and the number of pulverized cells (>20 breaks) calculated for 100 metaphases per sample.

Table 3.

Quantification and characterization of chromosome breakage in control and Fanconi Anemia (FA) blood samples after untreated and DEB-stimulated cultures

Parametera	Untreated		DEB 0.5 μg/mL
	Controls (N = 250)	FA (N = 88)	Controls (N = 250)	FA (N = 88)
Breaks and Fragments	0-3	0-66	0-21	7-1522
Radial Figures	0	0-8	0-4	3-226
Complex Radial Figures	0	0	0	1-42
Ring Chromosomes	0	0-2	0-1	0-18
Dicentric Chromosomes	0	0-3	0-2	0-8
Pulverized Cells	0	0	0	0-65

DEB, diepoxybutane.

^aDescriptive statistics are presented as the 2.5^th to 97.5^th percentile ranges per group, calculated for a number of 100 metaphases per case.

Figure 2.

Graphical distribution of CI parameters in spontaneous and DEB-treated cells. Boxplots representing the distribution of several CI parameters between the 2.5^th and 97.5^th percentiles, in both untreated cells and cells exposed to DEB at 0.5 μg/mL. Individual data points represent outlier values. Included parameters were (A-B) the percentage of aberrant cells, (C-D) percentage of multi-aberrant cells, (E-F) number of breaks per cell, (G-H) number of breaks and fragments, (I-J) number of radial figures, (K-L) number of ring chromosomes, (M-N) number of dicentric chromosomes, and (O) CFI. Complex radial figures and pulverized cells were not graphed, as untreated cells (both control and FA) and DEB-treated controls all had null values. CFI, chromosome fragility index; CI, chromosome instability; DEB, diepoxybutane.

In untreated cultures, both FA and control samples exhibited low overall levels of damage. Among controls, the number of breaks and fragments ranged from 0 to 3, with no radial figures or pulverized cells. FA cases showed a broader range of spontaneous instability, reaching up to 66 breaks and occasionally showing up to 8 simple radial figures, 2 rings, and 3 dicentric chromosomes. Despite these findings, none of the parameters differed significantly between the FA and control cases.

Upon DEB exposure (0.5 μg/mL), both the number and complexity of breaks increased significantly in FA cases. In controls, the number of breaks and fragments reached a maximum of 21, and the number of radial figures reached a maximum of 4, with no complex radials or pulverized cells observed. In contrast, FA cases exhibited extremely high levels of CI, with up to 1522 breaks and fragments, 226 radial figures, 42 complex radial figures, and 65 pulverized cells per 100 cells. In addition, ring and dicentric chromosomes were markedly increased in FA samples, ranging from 0 to 18 and 0 to 8 per case, respectively, whereas controls showed at most 1 ring and 2 dicentric chromosomes. These aberrations, while common in FA, are not specific to this disorder and are typically associated with other conditions involving telomere dysfunction.

In summary, the most important discriminative parameter was the presence of complex radial figures, which were absent in all control samples, reinforcing their diagnostic relevance as cytogenetic hallmarks of FA. The combined assessment of breaks per cell, CFI, and the presence of complex radial figures therefore provides optimal diagnostic accuracy.

Characterization of molecular findings in the FA cohort

Of the 93 cases, 50 (54%) had molecular confirmation of the diagnosis. Of these, 41 (82%) were homozygous and 9 (18%) were compound heterozygous. Consanguinity was reported in 36 of the 41 homozygous cases, of which 35 corresponded to Romani patients.

All cases without a molecular diagnosis were either not tested or lost to follow-up. With the exception of the previously described mosaic case, no other cases were classified as molecularly negative with a positive cytogenetic result. Of the 50 diagnoses, 45 (90%) were due to biallelic pathogenic variants in FANCA, and other represented genes were FANCD2 (HGNC:3585), FANCG, FANCJ/BRIP1 (HGNC:20473), and PALB2 (Tables 1 and 4; Supplemental Table 1). We observed all types of single-nucleotide variants (missense and presumably truncating), indels, and exon-level deletions, indicating a large set of potential FA-causing variants.

Table 4.

List and description of all genetic variants in molecularly confirmed Fanconi Anemia cases

Number of Cases	Gene	Variant 1 Description	Variant 1 Impact	Variant 1 Type	Variant 2 Description	Variant 2 Impact	Variant 2 Type	Allelism
35	FANCA	c.295C>T g.89811060G>A	p.(Gln99Ter)	Nonsense	-	-	-	Homozygous
1	FANCA	c.295C>T g.89811060G>A	p.(Gln99Ter)	Nonsense	seq[GRCh38] 16q24.3 (89808294_89811071)x1	Deletion of Exons 4-6	Copy-Number Loss	Compound Heterozygous
1	FANCA	c.718C>T g.89803333G>A	p.(Gln240Ter)	Nonsense	c.2870G>A g.89758688C>T	p.(Trp957Ter)	Nonsense	Compound Heterozygous
1	FANCA	c.1126C>T g.89792026G>A	p.(Gln376Ter)	Nonsense	c.(?_-1)_(79+1_80-1)del	Deletion of Exon 1	Copy-Number Loss	Compound Heterozygous
1	FANCA	c.1827-1G>T g.89775816C>A	r.spl	Splicing	c.3829-9G>A g.89740108C>T	r.spl	Splicing	Compound Heterozygous
1	FANCA	c.3556A>G g.89745029T>C	p.(Arg1186Gly)	Missense	c.3913C>T g.89740015G>A	p.(Leu1305Phe)	Missense	Compound Heterozygous
2	FANCA	c.3788_3790del g.89740849_89740851del	p.(Phe1263del)	Inframe Deletion	-	-	-	Homozygous
1	FANCA	c.3788_3790del g.89740849_89740851del	p.(Phe1263del)	Inframe Deletion	c.(1359+1_1360-1)_ _(1626+1_1627-1)del g.(89779958_89782858)_ _(89784965_89791402)del	Deletion of Exons 15-17	Copy-Number Loss	Compound Heterozygous
1	FANCA	c.3788_3790del g.89740849_89740851del	p.(Phe1263del)	Inframe Deletion	c.4199G>T g.89738943C>A	p.(Arg1400Leu)	Missense	Compound Heterozygous
1	FANCA	c.4069G>C g.89739231C>G	p.(Ala1357Pro)	Missense	arr[GRCh38] 16q24.3 (89733304_89843011)x1	Full Gene Deletion	Copy-Number Loss	Compound Heterozygous
1	FANCD2	c.376A>G g.10034797A>G	p.(Ser126Gly)	Missense	-	-	-	Homozygous
1	FANCD2	c.796_798del g.10042571_10042573del	p.(Val266del)	Inframe Deletion	c.2022-2A>G g.10064727A>G	r.spl	Splicing	Compound Heterozygous
1	FANCG	c.1077-2A>G g.35076030T>C	r.spl	Splicing	-	-	-	Homozygous
1	FANCJ/BRIP1	c.1341-3C>G g.61793732G>C	r.spl	Splicing	-	-	-	Homozygous
1	PALB2	c.1240C>T g.23635306G>A	p.(Arg414Ter)	Nonsense	-	-	-	Homozygous

Reference transcripts for the listed genes are NM_000135.4 (FANCA), NM_001018115.3 (FANCD2), NM_004629.2 (FANCG), NM_032043.3 (FANCJ/BRIP1), and NM_024675.4 (PALB2). Reference chromosomes for the listed genes are NC_000016.10 (FANCA), NC_000003.12 (FANCD2), NC_000009.12 (FANCG), NC_000017.11 (FANCJ/BRIP1), and NC_000016.10 (PALB2). All variant descriptors refer to the GRCh38 human genome version. The ACMG/AMP variant classification and criteria applied for each variant can be found in detail in Supplemental Table 1.

Of the 45 FANCA cases, 35 (78%) were caused by the homozygous NM_000135.4:c.295C>T p.(Gln99Ter) variant. This is a well-known founder variant in Iberian Romani, and all 35 patients were of this ethnicity. The remaining patient of Romani ethnicity, who corresponded to the prenatal case, was not homozygous for the founder variant. He was instead a compound heterozygote, with 1 allele having the c.295C>T variant and an allele bearing an intragenic FANCA deletion comprising exons 4 to 6 (which overlaps the c.295 position) (Table 4). While the mother was a heterozygote for the c.295C>T variant, we were unable to test the father. Therefore, we were unable to determine whether the intragenic deletion resulted from de novo recombinational errors or was inherited (something that has been observed multiple times for intragenic FANCA deletions²¹).

The causal variant was identified before cytogenetic testing for CI in 6 probands with a molecular diagnosis (12%), excluding those identified due to family history. All of these cases were diagnosed after 2017, likely reflecting the exponential increase in the availability of sequencing tests in clinical practice. Nevertheless, our findings reinforce the importance of functional CI testing for FA diagnosis, even in cases with prior molecular characterization, as it provides direct evidence of FA pathway dysfunction and remains particularly relevant for diagnostic confirmation and mosaicism assessment.

Clinical presentation and phenotypic assessment

We reviewed the clinical presentation of the 92 postnatal cases of FA at the time of diagnosis (Table 5). In 43 (47%) cases, there were both hematologic and morphologic phenotypes at diagnosis, while in 25 (27%) and 12 (13%) cases, the clinical presentation was only hematologic or morphologic, respectively. The remaining 12 (13%) patients were asymptomatic, and the diagnosis was performed as screening due to a positive family history in all cases. There was a positive family history in 5 additional cases, for a total of 17 (18%) in the cohort.

Table 5.

Characterization of clinical phenotypes of postnatal patients with FA at diagnosis

Variable	Descriptive Statistics (N = 93)
Hematologic features	67 (72%)
Thrombocytopenia	49 (53%)
Macrocytosis	38 (41%)
Anemia	36 (39%)
HbF Elevation	32 (34%)
Neutropenia	28 (30%)
Pancytopenia	21 (22%)
Morphologic features	55 (60%)
VACTERL-H Association	6 (7%)
Vertebral	2 (2%)
Anal	6 (7%)
Cardiopathy	13 (14%)
TEF	1 (1%)
Esophageal/Duodenal	8 (9%)
Renal	11 (12%)
Upper Limb	26 (28%)
Hydrocephalus	0
PHENOS Sign	3 (3%)
Pigmentation	30 (32%)
Small Head	15 (16%)
Small Eyes	17 (19%)
Neurostructural	3 (3%)
Otology	3 (3%)
Short Stature	24 (26%)

Descriptive statistics are presented as frequency (proportion).

FA, Fanconi anemia; HbF, fetal hemoglobin; TEF, tracheo-esophageal fistulae.

The most common cytopenia was thrombocytopenia, which was observed in over half of the cases, followed by anemia and neutropenia. In 21 (22%) patients, pancytopenia was present at diagnosis, reflecting trilinear bone marrow failure. The mean age of cases with this feature was 8.6 (± 6.5) years, which is consistent with previously reported data from other FA cohorts.¹⁰ Interestingly, we observed at least 1 morphologic phenotype in 18 (86%) patients with pancytopenia at diagnosis (Supplemental Table 2), which was significantly higher than in patients without pancytopenia (39 (53%) patients) (χ²1 = 7.124, P = .010, V = 0.275), indicating a more severe clinical presentation.

No single morphologic feature was observed in the majority of patients. The most common morphologic features in the overall cohort were the presence of hyperpigmented macules (30 (32%) patients), radial/thumb/upper limb anomalies (26 (28%) patients), and short stature (24 (26%) patients). The frequency of these features was similar to that of a previously reported FA review,^6,10,41 but was significantly lower than what was reported in recent cohorts.^6,41 This was also observed for the VACTERL-H association (presence of at least 3 of the following: Vertebral abnormalities, Anal atresia, Cardiac abnormalities, Tracheo-esophageal fistula, Esophageal or duodenal atresia, Renal abnormalities, upper Limb abnormalities, and Hydrocephalus) and PHENOS sign (presence of at least 4 of the following: skin Pigmentation abnormalities, small Head, small Eyes, structural central Nervous system abnormalities, Otologic abnormalities, and Short stature), which were observed in less than 10% of patients (Table 5). This could be attributed to the underascertainment of cases in retrospective cohorts when compared to the more recent longitudinal/prospective and cross-sectional studies. We also determined the number of singular morphologic features in each case and observed a median of 2 features (ranging from 0 to 7). In cases with at least 1 feature, the median number was 3 (ranging from 1 to 7). Finally, we observed developmental delay/intellectual disability in 14 (15%) cases.

We also assessed how the presence or absence of clinical phenotypes impacted the age at FA diagnosis. We observed that cases with hematologic features were diagnosed significantly earlier (5.7 ± 8.9 years) than those without hematologic features (10.1 ± 7.8 years) (t(88) = 2.260, p = 0.026, g = 0.541). A correlation between the values of several hematologic parameters and age at diagnosis was also performed (Supplemental Table 3). Significant negative correlations between white blood cell counts and age at diagnosis, and positive correlations between mean corpuscular volume and age were observed. This suggests that leukopenia and macrocytosis may lead clinicians to pursue the FA diagnosis earlier. Finally, we also observed that cases with at least 1 morphologic feature were diagnosed earlier (7.4 ± 6.9 years) than those with none (11.4 ± 9.6 years) (t(88) = 2.344, p = 0.021, g = 0.503). No significant correlation was observed between age at diagnosis and the number of individual morphologic features in patients (ρ = −0.121, p = 0.127).

As FA predisposes to different cancer types, we reviewed the database for oncological complications. Out of the 92 postnatal cases, 9 (10%) had a history of cancer, namely 6 with acute myelogenous leukemia (median age at diagnosis of 14 years [3-28]); 1 with medulloblastoma at 2 years of age; 1 with vulvar squamous cell carcinoma (SCC) at 36 years of age; and 1 with 3 primary malignancies: invasive breast cancer at 34, anal SCC at 35, and vulvar SCC at 39 years of age. Regarding the genotypes of these 9 cases, 2 of the patients with acute myelogenous leukemia did not have a molecular diagnosis; the patient with medulloblastoma had the PALB2 variant; all other patients had FANCA biallelic variants. No cases of head and neck SCC were identified. These complications are likely underascertained, given that the median age at last follow-up in our database is 13 years (range: 0-39 years).³⁹ The single case in our cohort with a prenatal FA diagnosis was a fetus with several malformations identified in a morphologic ultrasound assessment during the 2nd trimester of pregnancy. The parents were of Romani ethnicity and were once-removed first-degree cousins. Fetal assessment revealed duodenal atresia, horseshoe kidney, and atrioventricular discordance. Pregnancy was terminated at 21 weeks due to the severe malformations, and fetal autopsy additionally revealed absence of lung fissures, proximally placed thumbs, and wrist arthrogryposis. This fulfilled the criteria for the VACTERL-H association.

Characterization of the Romani cohort of FANCA c.295C>T homozygous patients

As previously stated, our cohort comprises 35 postnatal FA cases of Romani patients, all of whom had molecular confirmation of the FANCA c.295C>T variant in homozygosity. This variant presumably creates a premature stop codon and leads to a nonsense change, p.(Gln99∗).²¹ The most commonly observed phenotypes in this subgroup are described in Supplemental Table 4.

We started by comparing the overall clinical presentation between cases of Romani versus all other ethnicities from our cohort. While the frequency distribution was similar for cases presenting with morphologic-only or both hematologic and morphologic phenotypes, Romani patients presented significantly more frequently with no phenotype and less frequently with hematologic-only phenotypes (Table 6). Since cases presenting with no phenotype are typically diagnosed due to a positive family history, we observed, as expected, that Romani cases had a significantly higher rate of positive family history (12 (34%)) than those of other ethnicities (5 (9%)) (χ²(1) = 9.879, P = .002, V = 0.324). No specific hematologic phenotype had significantly different frequencies between Romani and other ethnicities (Supplemental Table 4). For morphologic phenotypes, short stature and microcephaly were more common in Romani patients (Supplemental Table 4).

Table 6.

Comparison of the overall clinical presentation in Romani vs other ethnicities FA cases

Clinical Presentation	Romani/ FANCA c.295C>T (N = 35)	All Non-Romani Cases (N = 57)
		Descriptive	Comparison
None	8 (23%)	4 (7%)	χ2 (3) = 13.678 P = .003 V = 0.386
Hematologic	4 (11%)	21 (7%)
Morphologic	8 (23%)	4 (7%)
Both	15 (43%)	28 (49%)

FA, Fanconi anemia.

Correlations between chromosome instability and clinical presentation

Next, we wanted to assess whether CI had any relation with the clinical phenotype. We started by comparing the number of chromosome breaks per cell between different hematologic phenotypes. We observed a significantly increased number of breaks per cell in patients presenting with pancytopenia (Figure 3A) and anemia (isolated or not) (Figure 3B). No differences were observed for patients with or without neutropenia (Figure 3C) and thrombocytopenia (Figure 3D).

Figure 3.

Correlation between chromosome instability and hematologic FA phenotype. (A) Comparison of the mean breaks per cell between patients with and without pancytopenia, (B) anemia, (C) neutropenia, or (D) thrombocytopenia at diagnosis. (E) Correlations between mean breaks per cell and the levels of hemoglobin, (F) red blood cells (RBC) content, (G) neutrophils, and (H) platelets at diagnosis. FA, Fanconi anemia; RBC, red blood cells.

When correlating hematologic parameters with the number of breaks per cell, we observed a significant negative correlation of moderate strength with both the hemoglobin concentration (Figure 3E) and red blood cell (RBC) content (Figure 3F). On the other hand, no significant correlation was observed for the neutrophil (Figure 3G) or platelet number (Figure 3H). Correlations with other hematologic parameters were nonsignificant (Supplemental Table 5).

Regarding morphologic features, we did not observe a significant difference in the number of breaks per cell between patients with (6.7 ± 3.8 breaks per cell) and without (6.6 ± 4.4 breaks per cell) at least one of these features (t(87) = 0.095, P = .924, g = 0.021). Accordingly, we did not observe a significant correlation between breaks per cell and the number of morphologic features each patient presented (ρ = −0.062, P = .565).

Overall, these data indicate that increased CI may be associated with a more severe hematologic phenotype, predominantly related to the RBC lineage.

Discussion

In this study, we present data regarding the Portuguese cohort of patients with FA, comprising the long-term diagnostic experience of an expert cytogenetics laboratory and the largest data set of patients with FA at a national level. We present relevant findings regarding both the clinical presentation and follow-up, as well as important considerations on the diagnostic assessment, specifically in cytogenetic methods. An assessment of the correlation between CI and clinical presentation was also performed.

Overall, the clinical presentation (in terms of phenotypes and their frequency) is similar to the reported one based on the aggregation of data from different studies.¹⁰ However, other large national cohorts, such as the Turkish cohort with 175 cases⁴² and the Chinese cohort with 148 cases,⁴³ reported different phenotypic frequencies, which may reflect different genetic variation backgrounds and also different rates of consanguinity. In the Turkish cohort,²⁸ the rate of hematologic changes at diagnosis was similar to what we observed (75% and 72%, respectively). On the other hand, rates of microcephaly (93%), hyperpigmentation (70 to 80%), and radial/thumb anomalies (53%) were much higher than what we observed.¹⁰ The rate of consanguinity in this Turkish cohort was 73%, which is significantly higher than what we observed in our Portuguese cohort (41%), and may explain the underlying differences in clinical presentation. Our method of collecting clinical data can also underestimate the presence of specific morphologic features, as it was based on the review of clinical records. Furthermore, no patient from that cohort had a molecular diagnosis and may represent a genetically distinct sample in terms of FA-causal variants. In the Chinese cohort,²⁹ overall rates of malformation were also higher, yet individual features presented a similar frequency to what we observed, with dermatological and radial/thumb anomalies being present in 25 (30%) of the cases.

It is also noteworthy that about one third of our cohort is represented by a specific population with a distinct genetic background and high rates of consanguinity.⁴⁴ Differences observed in the clinical presentation of this subset of patients from the remainder of the cohort can be attributed to their distinct founder FANCA c.295A>T variant (originally described in the Spanish Romani population⁴⁵) and/or to their additional differences in genetic background compared to the European non-Romani population. It is also noteworthy that the low proportion of patients fulfilling the VACTERL-H association or PHENOS sign may be attributed to the higher representation of FANCA cases compared to other cohorts, as the presence of either is inversely correlated with this genotype.⁶ One additional important aspect is that there was a significantly higher rate of diagnosis of asymptomatic patients through a positive family history, which further supports the relevance of providing personalized genetic counseling regarding the patient’s ethnic background.

We also presented one of the largest cohorts with a detailed cytogenetic assessment of patients with FA. Quantification of DEB-induced chromosome breaks is a sensitive and precise method for FA diagnosis. In our cohort, almost no overlap was observed between control and FA samples when breaks per cell were quantified, with the exception of cases that were likely mosaics. This suggests that potential cases with strong clinical suspicion for FA and a CI assessment below the predefined cutoffs should motivate additional analysis toward FA suspicion, which cannot be unequivocally excluded.

In parallel with DEB-induced CI, quantification of spontaneous CI was also performed. Although FA samples tended to show higher levels of spontaneous CI, there was a clear overlap between the 2 populations, indicating that spontaneous CI cannot be used as a reliable discriminative parameter for diagnosis. Instead, it serves as a reference for basal CI prior to DEB exposure.

An important outcome regarding the CI results can be noted. We observed that the 15 cases with follow-up CI evaluation revealed substantial intra-individual variability in the number of chromosome breaks per cell, with no consistent pattern over time. These variations are likely attributable to clinical, biological, and technical factors, such as variability in hemogram values, the influence of oxidative stress, and an increased number of pulverized cells, which may influence the final calculation of the number of breaks per cell. Among these cases, 2 had a reversal of cellular phenotype in peripheral blood, with CI values decreasing below the diagnostic cutoff for FA during follow-up. This phenomenon has been previously observed in FA and other monogenic disorders of the hematopoietic system.^40,46 Notably, both patients had been diagnosed in childhood in a familial FA context and had clearly positive CI results at diagnosis. Despite the subsequent normalization of CI in peripheral blood, clinical features within the FA spectrum were present in both cases, including nonhematologic manifestations. Neither patient had any hematologic abnormalities at diagnosis or at follow-up, the latest of which was at 38 years of age in one of the patients. These findings suggest that, in FA cases with no hematologic abnormalities or with hematologic improvement, cytogenetic reassessment may have prognostic value, as spontaneous somatic reversal can occur. Such information may contribute to a more refined risk stratification and help with clinical decision-making regarding the periodicity of bone marrow failure surveillance, particularly considering the invasive nature of bone marrow evaluations.

An assessment of the correlation of CI with clinical phenotype has been recently published in a cohort of Spanish patients.³² However, as previously stated, a low/high cutoff for CI was used for grouped assessment, which was predetermined based on the internal sample’s median breakage in FA cases. Here, we present direct correlations between the levels of DEB-induced breaks per cell and different phenotypes and analytic markers. We observed that patients with morphologic features did not present significantly different breakage levels. However, cases with phenotypes suggestive of bone marrow failure did present with increased mean breaks per cell. This finding was predominantly related to RBCs, as increased breakage was detected in patients with lower levels of hemoglobin and RBC content. This suggests that breakage levels are a better diagnostic/prognostic marker for hematologic phenotypes than morphologic ones, which agrees with the fact that CI was assessed specifically in hematopoietic tissue. This is in line with the findings from the Spanish cohort, in which the “high fragility” group presented with a higher cumulative incidence of hematologic phenotypes.³² Interestingly, these findings are in line with previous observations suggesting that FA RBCs already exist in a basal, preconditioned oxidative stress state and progressively fail to maintain an adequate stress response under increased oxidative and genotoxic challenges. This impaired adaptive capacity may contribute to increased cellular vulnerability and chromosome damage, potentially explaining the association between increased breakage levels and altered RBC parameters.⁴⁷

One of the main limitations of this study is that it represents a single population, and its findings may not be generalizable to the broader FA community. Nevertheless, accurate and objective descriptions of specific disease populations contribute significantly to the overall understanding of the disease. Regarding FA incidence, it is also noteworthy that our proposed estimate may be regarded as a minimum approximation of the real incidence because of the incomplete ascertainment of cases. In addition, based on our demographic data, we were only able to provide an estimate of the overall incidence, since ethnicity-specific incidence cannot be reliably calculated owing to the lack of specific demographic data.

Another limitation is that no genotype-phenotype correlations can be established other than for the specific Romani founder variant, as for most variants there is only one representative case. This is a common limitation in most FA studies, even when including large patient cohorts. All correlations between molecular/cytogenetic and clinical data should also be interpreted cautiously, as they derive from static data, although they represent a test for the association of features at a simultaneous time point. Finally, and as previously stated, we also recognize that reviewing electronic health records for clinical data extraction can often lead to an underestimation of different findings that might have been present and assessed but were not recorded.

In conclusion, we described and characterized the largest Portuguese cohort of FA to date. We provided clinical and demographic data that are very relevant for the specific characterization of a national population. Nevertheless, this cohort has provided us with valuable data regarding the importance of follow-up cytogenetic assessment of FA cases and the direct correlation between CI and RBCs.